The Shaping of Modern Human Immune Systems by Multiregional

Admixture with Archaic Humans

Laurent Abi-Rached et al.
Science 334, 89 (2011);
DOI: 10.1126/science.1209202

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your
colleagues, clients, or customers by clicking here.
Permission to republish or repurpose articles or portions of articles can be obtained by
following the guidelines here.

The following resources related to this article are available online at

Downloaded from www.sciencemag.org on February 28, 2013

www.sciencemag.org (this information is current as of February 28, 2013 ):

Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
http://www.sciencemag.org/content/334/6052/89.full.html
Supporting Online Material can be found at:
http://www.sciencemag.org/content/suppl/2011/08/19/science.1209202.DC1.html
A list of selected additional articles on the Science Web sites related to this article can be
found at:
http://www.sciencemag.org/content/334/6052/89.full.html#related
This article cites 30 articles, 13 of which can be accessed free:
http://www.sciencemag.org/content/334/6052/89.full.html#ref-list-1
This article has been cited by 11 articles hosted by HighWire Press; see:
http://www.sciencemag.org/content/334/6052/89.full.html#related-urls
This article appears in the following subject collections:
Genetics
http://www.sciencemag.org/cgi/collection/genetics

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2011 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.

climate differentiation and in the top 5% of glob-
al iEH (Table 1). This screen identified four SNPs
within four candidate genes, and linked within
10 kbp to eight additional candidate genes. The
three most interesting candidate genes with the
associated SNP situated in the coding region
were the LAC1 gene (AT1G18140), involved in
the response to necrotrophs (21), associated with
silique number in Germany; CHR8 (AT2G18770),
involved in DNA repair after viral infection (22),
associated with survival in Germany; and SAG21
(AT4G02380), involved in water stress tolerance
(23, 24) and associated with survival in Finland.
Note that for both CHR8 and SAG21, the allele
showing iEH evidence of a selective sweep cor-
responds to the minor and geographically re-
stricted allele. The CHR8 allele associated with
low survival in Germany is established in the
over, alleles associated with high fitness within
sites tend to be local alleles linked to partic-
ular climatic factors, providing evidence of lo-
cal adaptation in A. thaliana at the scale of the
European continent. Our finding that loci as-
sociated with fitness in natural environments
display geographic and climatic signatures of
adaptation complements recent observations
of candidate loci associated with climate (10).
GWAS of fitness traits measured in multiple
natural environments, combined with geograph-
ic and climatic analyses, constitutes a powerful
approach to identify environment-specific can-
didate genes for local adaptation.
References and Notes
1. J. Hereford, Am. Nat. 173, 579 (2009).
2. S. N. Aitken, S. Yeaman, J. A. Holliday, T. L. Wang,
S. Curtis-McLane, Evol. Appl. 1, 95 (2008).
REPORTS
15. H. M. Kang et al., Nat. Genet. 42, 348 (2010).
16. See supporting material on Science Online.
17. S. J. Phillips, R. P. Anderson, R. E. Schapire, Ecol. Modell.
190, 231 (2006).
18. P. C. Sabeti et al., Nature 419, 832 (2002).
19. K. Tang, K. R. Thornton, M. Stoneking, PLoS Biol. 5, e171
(2007).
20. D. De Carvalho et al., Mol. Ecol. 19, 1638 (2010).
21. L. Pourcel et al., Plant Cell 17, 2966 (2005).
22. J. Molinier, E. J. Oakeley, O. Niederhauser, I. Kovalchuk,
B. Hohn, Mutat. Res. 571, 235 (2005).
23. L. M. Weaver, S. S. Gan, B. Quirino, R. M. Amasino,
Plant Mol. Biol. 37, 455 (1998).
24. S. B. Mowla et al., Plant J. 48, 743 (2006).
25. A. C. McCormac, G. C. Whitelam, M. T. Boylan, P. H. Quail,
H. Smith, J. Plant Physiol. 140, 707 (1992).
Acknowledgments: We thank M. Blzquez, C. Dean,
M. Hoffmann, H. Kuittinen, and O. Savolainen for
hosting the experiments; J. Anderson, D. Eaton, J. F. Egan,
C. Lopez-Gallego, L. J. Martin, C. D. Muir, R. Petipas, J. Roe,
and R. N. Schaeffer for managing the field experiments;
Brown undergraduates for fruit counts; and J. Bergelson,

Downloaded from www.sciencemag.org on February 28, 2013

species northern range, and the SAG21 allele
associated with low survival in Finland is re-
stricted to the southern part of the species range
(Fig. 3). Recent positive selection may have fa-
vored these alleles locally, but their geographic
range may be restricted by negative pleiotropic
effects on fitness in other environments, such as
our field sites.
This study provides robust molecular evidence
for broad scale local adaptation in A. thaliana.
By investigating SNPs linked to loci experienc-
ing real-time selection in different natural en-
vironments, we found that the genetic basis of
fitness differs dramatically across sites. More-
3. J. R. Bridle, T. H. Vines, Trends Ecol. Evol. 22, 140
(2007).
4. A. S. Jump, J. Peuelas, Ecol. Lett. 8, 1010 (2005).
5. D. H. Reed, R. Frankham, Evolution 55, 1095
(2001).
6. I. M. Ehrenreich, M. D. Purugganan, Am. J. Bot. 93,
953 (2006).
7. J. Bergelson, F. Roux, Nat. Rev. Genet. 11, 867
(2010).
8. J. Stapley et al., Trends Ecol. Evol. 25, 705 (2010).
9. I. Baxter et al., PLoS Genet. 6, e1001193 (2010).
10. A. J. Eckert et al., Mol. Ecol. 19, 3789 (2010).
11. M. C. Hall, D. B. Lowry, J. H. Willis, Mol. Ecol. 19, 2739
(2010).
12. C. Mariac et al., Mol. Ecol. 20, 80 (2011).
13. A. M. Wilczek et al., Science 323, 930 (2009).
14. S. Atwell et al., Nature 465, 627 (2010).
A. Hancock, and J. Mossman for comments on the
manuscript. Supported by NSF grant EF-0425759
and an Alexander von Humboldt Research Award.
Supplemental data for the phenotypes and the associated
SNPs are available at http://dx.doi.org/10.5061/
dryad.37f9t. Genotype data for the SNPs used in this study
are available at http://regmap.uchicago.edu.
Supporting Online Material
www.sciencemag.org/cgi/content/full/334/6052/86/DC1
Materials and Methods
Figs. S1 to S7
Tables S1 to S6
References
2 June 2011; accepted 25 August 2011
10.1126/science.1209271

for such admixture. Because of their vital func-

The Shaping of Modern Human tions in immune defense and reproduction, as
ligands for T cell and natural killer (NK) cell re-
Immune Systems by Multiregional 1
Department of Structural Biology and Department of Mi-
Admixture with Archaic Humans crobiology and Immunology, Stanford University School of
Medicine, Stanford, CA 94305, USA. 2Department of An-
thropology, Santa Clara University, Santa Clara, 95050, USA.
Laurent Abi-Rached,1 Matthew J. Jobin,2,3 Subhash Kulkarni,1 Alasdair McWhinnie,4 Klara Dalva,5 3
Department of Anthropology, Stanford University, Stanford,
Loren Gragert,6 Farbod Babrzadeh,7 Baback Gharizadeh,7 Ma Luo,8,9 Francis A. Plummer,8,9 CA 94305, USA. 4Anthony Nolan Research Institute, Royal Free
Hospital, London NW3 2QG, UK. 5Department of Hematology,
Joshua Kimani,10 Mary Carrington,11,12 Derek Middleton,13 Raja Rajalingam,14 Meral Beksac,5
Ankara University, 06520 Ankara, Turkey. 6National Marrow
Steven G. E. Marsh,4,15 Martin Maiers,6 Lisbeth A. Guethlein,1 Sofia Tavoularis,16 Donor Program, Minneapolis, MN 55413, USA. 7Stanford Ge-
Ann-Margaret Little,4,15 Richard E. Green,17 Paul J. Norman,1 Peter Parham1* nome Technology Center, Stanford University School of Medi-
cine, Palo Alto, CA 94304, USA. 8Public Health Agency of
Canada, National Microbiology Laboratory, Winnipeg, Manitoba
Whole genome comparisons identified introgression from archaic to modern humans. Our analysis of R3E 3R2, Canada. 9Department of Medical Microbiology, Uni-
highly polymorphic human leukocyte antigen (HLA) class I, vital immune system components subject to versity of Manitoba, Winnipeg, Manitoba R3E 0J9, Canada.
strong balancing selection, shows how modern humans acquired the HLA-B*73 allele in west Asia through 10
Department of Medical Microbiology, University of Nairobi,
admixture with archaic humans called Denisovans, a likely sister group to the Neandertals. Virtual Nairobi 00202, Kenya. 11Cancer and Inflammation Program,
Laboratory of Experimental Immunology, SAIC-Frederick, Inc.,
genotyping of Denisovan and Neandertal genomes identified archaic HLA haplotypes carrying functionally National Cancer Institute-Frederick, Frederick, MD 21702,
distinctive alleles that have introgressed into modern Eurasian and Oceanian populations. These alleles, USA. 12Ragon Institute of MGH, MIT and Harvard, Boston, MA
of which several encode unique or strong ligands for natural killer cell receptors, now represent more than 02129, USA. 13Division of Immunology, School of Infection
half the HLA alleles of modern Eurasians and also appear to have been later introduced into Africans. and Host Defense, University of Liverpool, Transplant Immu-
nology, Royal Liverpool University Hospital, Liverpool L7 8XP,
Thus, adaptive introgression of archaic alleles has significantly shaped modern human immune systems.
UK. 14UCLA Immunogenetics Center, Department of Pathology
and Laboratory Medicine, David Geffen School of Medicine at
hether or not interbreeding occurred genomes (4). These studies, based on statistical

W
between archaic and modern humans
has long been debated (1, 2). Recent
estimates suggest that Neandertals contributed
1 to 4% to modern Eurasian genomes (3), and
Denisovans, a likely sister group to the Neander-
tals, contributed 4 to 6% to modern Melanesian
genome-wide comparisons, did not address if
there was selected introgression of functionally
advantageous genes (5). We explored whether the
highly polymorphic HLA class I genes (HLA-A,
-B, and -C) (fig. S1) of the human major histo-
compatibility complex (MHC) are sensitive probes
UCLA, University of California at Los Angeles, Los Angeles, CA
90095, USA. 15UCL Cancer Institute, University College London,
Royal Free Campus, London WC1E 6BT, UK. 16Canadian Blood
Services, Head Office, HLA Laboratory, Ottawa, Ontario KIG 4J5,
Canada. 17Department of Biomolecular Engineering, University
of California Santa Cruz, Santa Cruz, CA 95064, USA.
*To whom correspondence should be addressed. E-mail:
peropa@stanford.edu

www.sciencemag.org SCIENCE VOL 334 7 OCTOBER 2011 89

REPORTS
ceptors, maintaining a variety of HLA-A, -B, and
-C proteins is critical for long-term human sur-
vival (6). Thus, HLA-A, -B, and -C are subject to
strong multiallelic balancing selection, which,
with recombination, imbues human populations
with diverse HLA alleles and haplotypes of dis-
tinctive structures and frequencies (7).
An exceptionally divergent HLA-B allele is
HLA-B*73:01 (8, 9). Comparison with the other
>2000 (10) HLA-B alleles and chimpanzee and
gorilla alleles from the same locus (MHC-B)
shows that HLA-B*73:01 is most closely related
to subsets of chimpanzee and gorilla MHC-B
alleles (11) (figs. S2 to S4). This relation extends
throughout a ~9-kilobase (kb) region of the
B*73:01 haplotype (Fig. 1A) and defines a deep-
ly divergent allelic lineage (MHC-BII ), distinct
from the MHC-BI lineage to which other human
~16 million years ago (Fig. 1B), well before the
split between humans and gorillas, but whereas
MHC-BI comprises numerous types and subtypes,
MHC-BII is only represented in modern humans
by B*73:01 (fig. S5). HLA-B*73:01 combines an-
cient sequence divergence with modern sequence
homogeneity, properties compatible with modern
humans having recently acquired HLA-B*73:01
through introgression.
In modern humans, HLA-B*73 is concentrated
in west Asia and is rare or absent in other regions
(12) (Fig. 1C and fig. S6). This distribution is
consistent with introgression of HLA-B*73 in
west Asia, a site of admixture between modern
and archaic humans (3). Also consistent with in-
trogression is the linkage disequilibrium (LD) be-
tween B*73:01 and HLA-C*15:05 (13), an allele
having wider distribution than B*73, but con-
Worldwide, ~98% of people carrying B*73
also carry C*15:05 (Fig. 1E and fig. S7). In Af-
ricans, the LD reaches 100%, but in west Asians,
it is weaker (~90%). These data are all consist-
ent with introgression in west Asia of an archaic
B*73:01-C*15:05 haplotype that expanded in
frequency there, before spreading to Africa and
elsewhere. HLA-B*73 is absent from Khoisan-
speaking and pygmy populations who likely
diverged from other Africans before the Out-of-
Africa migration (14) (fig. S8). That Khoisan
and pygmies uniquely retain ancient mitochon-
drial and Y-chromosome lineages (14, 15), as
well as MHC-BI diversity (fig. S8), suggests that
B*73 was probably not present in any African
population at the time of the migration. These
data argue for models in which modern humans
acquired B*73 by archaic admixture in west

Downloaded from www.sciencemag.org on February 28, 2013

HLA-B alleles belong. These two lineages diverged centrated in west and southeast Asia (Fig. 1D). Asia and against models in which B*73 arose in

A E1 E2 E3 HLA-B*73 E4 E5 E6-7 E
B*73 + individuals
6 8 9 10
7
Associated
HLA-C alleles
1 2 3 4 5 11 12 13 14 (%)
Geographic
1 10kb 20kb 30kb 38kb N
region C*15 Not C*15
B Gene loss Gorilla
not not
15:05 12:02
15:05 12:02
MHC-B MHC-BII
Gene Chimpanzee Europe 2,677 98.4 0.3 0.4 0.9
duplication Inter-locus
16.3 MYA recombination Europe# 2,907 98.5 0.3 0.3 0.9
[13.8-19.9 MYA]
Africa 39 100 0.0 0.0 0.0
Human Africa## 90 97.8 2.2 0.0 0.0
HLA-B*73 W Asia 128 89.8 5.5 0.8 3.9
Formation of MHC-BI
two lineages Preservation of two lineages at the same locus N/S/E Asia 53 92.5 5.7 1.9 0.0
C D Other 498 99.0 0.0 0.4 0.6
Total 3,676 98.2 0.5 0.4 0.9

F
Modern/archaic homo split
270-440kya Models
B*73+ B*73+ a b
C*15- C*15-
ARCHA
IC
MODERN

4.9 Out-of-Africa migration

4.5
4.0 4.0 67.5kya
3.0 3.0
af
2.0 (%) 2.0 af
(%) Admixture
1.0 1.0 Emergence
0.0 0.0 of C*15:05 50kya
65kya
HLA-B*73:01 HLA-C*15:05

Back-to-Africa Disappear
Fig. 1. Modern humans acquired HLA-B*73 from archaic humans. (A) The B*73 haplotype contains B*73+
migration C*15+ 30kya
segments most closely related to chimpanzee and gorilla MHC-B alleles (green) and flanking segments 10kya

highly related to other HLA-B (blue) (brown segment is related to HLA-C) (fig. S4). (B) B*73s divergent B*73-
C*15+
core has its roots in a gene duplication that occurred >16 million years ago (MYA on figure). (Left to
right) MHC-B duplicated and diverged to form the MHC-BI and BII loci. One allele of BII recombined to Africa West Asia Eurasia

the BI locus, giving rise to the ancestor of B*73 and its gorilla and chimpanzee equivalents. B*73 is
thus the only remnant in modern humans of a deeply divergent allelic lineage. Mean and 95% credibility interval. (C to E) B*73:01 is predominantly found
outside Africa (C) as is C*15:05 (D), which is strongly associated with B*73 in 3676 individuals worldwide (E). (C and D) Color scale bars give allele frequency
(af) categories (top number, highest tick mark). Individuals with the B*73 haplotype were categorized on the basis of their geographic origin and the status of
the most commonly linked (C*15) and second-most commonly linked (C*12:02) HLA-C alleles (fig. S24). Number sign (#) includes Hispanic-Americans, double
number sign (##) includes African-Americans. (F) Archaic admixture (model a) or African origin (model b) could explain the distribution and association of
B*73 with C*15:05; simulations favor the former (a = 0.01 to 0.001) (figs. S9 to S11) (11). The large dotted box indicates the part of the models examined
by simulation; kya, thousand years ago.

90 7 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org


Africa and was carried to other continents in the
Out-of-Africa migration (Fig. 1F), as do the re-
sults of coalescence simulations that implement
rejection-based approximate Bayesian inference
(16) (a = 0.01 to 0.001) (figs. S9 to S11).
By reanalyzing genomic sequence data (3, 4, 11),
we characterized archaic HLA class I content from a
Denisovan and three Neandertals. The Denisovans
two HLA-A and two HLA-C allotypes are identi-
cal to common modern allotypes, whereas one
HLA-B allotype corresponds to a rare modern
recombinant allotype, and the other has never
been seen in modern humans (Fig. 2B and fig.
S12). The Denisovans HLA type is thus consist-
ent with an archaic origin and the known propen-
sity for HLA-B to evolve faster than HLA-A and
HLA-C (17, 18).
Not knowing the haplotype phase, we ex-
amined all possible combinations of Denisovan
HLA-A and HLA-C for their current distribution
worldwide. All four combinations are present in
Asia and Oceania, but absent from sub-Saharan
Africa, and uncommon in Europe (Fig. 2, C and D,
and fig. S13). Genome-wide comparisons showed
that modern and archaic non-African genomes
share only 10 long, deeply divergent haplotypes (3),
which are all considerably shorter (100 to 160 kb)
than the ~1.3-megabase (Mb) HLA-A-C haplotype
REPORTS
(Fig. 2A). Because modern HLA haplotypes di-
versify rapidly by recombination (1719), it is im-
probable that the HLA-A-C haplotypes shared
by modern humans and Denisovans were pre-
served on both lines since modern and Denisovan
ancestors separated >250 thousand years ago
(kya) [~10,000 generations (4)]. More likely is
that modern humans acquired these haplotypes
by recent introgression from Denisovans [note
II.6 in (11)]. Both alternative haplotype pairs are
common in Melanesians and reach 20% frequen-
cy in Papua New Guinea (PNG), consistent with
genome-wide assessment of Denisovan admix-
ture in Melanesians (4). The current distribution

Downloaded from www.sciencemag.org on February 28, 2013

Fig. 2. Effect of adaptive introgression of Denisovan
HLA class I alleles on modern Asian and Oceanian
populations. (A) Simplified map of the HLA class I
region showing the positions of the HLA-A, -B,
and -C genes. (B) Five of the six Denisovan HLA-A,
-B, and -C alleles are identical to modern counter-
parts. Shown at the left for each allele is the num-
ber of sequence reads (4) specific to that allele and
their coverage of the ~3.5-kb HLA class I gene. Cen-
ter columns give the modern-human allele (HLA type)
that has the lowest number of single-nucleotide
polymorphism (SNP) mismatches to the Denisovan
allele. The next most similar modern allele and the
number of SNP differences are shown in the columns on the right. A recombinant allele with 5 segments originating from B*40. The coding sequence is
identical to C*15:05:02. (C and D) Worldwide distributions of the two possible Denisovan HLA-A to -C haplotype combinations. Both are present in modern
Asians and Oceanians but absent from sub-Saharan Africans. (E to G) The distribution of three Denisovan alleles: HLA-A*11 (E), C*15 (F), and C*12:02 (G), in
modern human populations shows they are common in Asians but absent or rare in sub-Saharan Africans. (H) Estimation of divergence times shows that A*11,
C*15, and C*12:02 were formed before the Out-of-Africa migration. Shown on the left are the alleles they diverged from, on the right are the divergence time
estimates: median, mean, and range.

www.sciencemag.org SCIENCE VOL 334 7 OCTOBER 2011 91

REPORTS
of the Denisovan haplotypes (Fig. 2, C and D,
and fig. S13) shows, however, that Denisovan ad-
mixture widely influenced the HLA system of
Asians and Amerindians.
Of the two Denisovan HLA-A alleles (Fig. 2B),
A*02 is widespread in modern humans, whereas
A*11 is characteristically found in Asians (Fig.
2E) and reaches 50 to 60% frequency in PNG and
China, is less common in Europe, and is absent
from Africa (fig. S14). This distribution cou-
pled with the sharing of long HLA-A-C haplo-
types between Denisovans and modern Asians,
particularly Papuans (fig. S13), indicates that
Denisovan admixture minimally contributed the
A*11:01-C*12 or A*11:01-C*15 haplotype to mod-
ern Asians. A*11:01, which is carried by both these
archaic haplotypes, is by far the most common
A*11 subtype (12). Because HLA alleles evolve
dently in Denisovan and modern humans through-
out >250,000 years (4), as would be required if the
Out-of-Africa migration contributed any signif-
icant amount of A*11. The more parsimonious
interpretation is that all modern A*11 is derived
from Denisovan A*11 and that, after introgres-
sion, it increased in frequency to ~20% to become
almost as common in Asia as A*02 at ~24% (11).
Denisovan HLA-C*15 and HLA-C*12:02 are
also characteristic alleles of modern Asian popula-
tions (Fig. 2, F and G, and fig. S14). At high fre-
quency in PNG, their distribution in continental
Asia extends further west than A*11 does, and in
Africa, their frequencies are low. C*12:02 and C*15
were formed before the Out-of-Africa migration
(Fig. 2H and fig. S15) and exhibit much higher
haplotype diversity in Asia than in Africa (fig. S16),
contrasting with the usually higher African genetic
humans through admixture with Denisovans in
west Asia, with later spreading to Africa (21, 22)
(Fig. 1F and fig. S11 for C*15). Given our min-
imal sampling of the Denisovan population, it is
remarkable that C*15:05 and C*12:02 are the
two modern HLA-C alleles in strongest LD with
B*73 (Fig. 1E). Although B*73 was not carried by
the Denisovan individual studied, the presence of
these two associated HLA-C alleles provides strong
circumstantial evidence that B*73 was passed
from Denisovans to modern humans.
Genome-wide analysis showing that three
Vindija Neandertals exhibited limited genetic di-
versity (3) is reflected in our HLA analysis: Each
individual has the same HLA class I alleles (fig.
S17). Because these HLA identities could not be
the consequence of modern human DNA con-
tamination of Neandertal samples, which is <1%

Downloaded from www.sciencemag.org on February 28, 2013

subtype diversity rapidly (17, 18), it is highly im- diversity (20). These properties fit with C*12:02 (3), they indicate that these individuals likely
probable that A*11:01 was preserved indepen- and C*15 having been introduced to modern belonged to a small and isolated population (fig.

Fig. 3. Effect of adaptive introgression of Nean-

dertal HLA class I alleles on modern human pop-
ulations. (A) All six Neandertal HLA-A, -B, and -C
alleles are identical to modern HLA class I alleles.
Shown at the left for each allele is the number of
allele-specific sequence reads (3) and their cover-
age of the ~3.5-kb HLA gene. Center columns give
the modern-human allele (HLA type) having the
lowest number of SNP differences from the Nean-
dertal allele. The next most similar modern allele
and the number of SNP differences are shown in
the columns on the right. Alleles marked with include additional rare alleles. (B and C) Worldwide distributions of the two possible Neandertal HLA-A to -C
haplotype combinations. Both are present in modern Eurasians, but absent from sub-Saharan Africans. (D to G) Distribution of four Neandertal alleles: HLA-
B*07:02/03/06 (D), B*51:01/08 (E), C*07:02 (F), and C*16:02 (G), in modern human populations.

92 7 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org


S18). Clearly identified in each individual were
HLA-A*02, C*07:02, and C*16; pooling the three
sequence data sets allowed identification of HLA-
B*07, -B*51 and either HLA-A*26 or its close
relative A*66 as the other alleles (Fig. 3A). As
done for the Denisovan, we examined all combi-
nations of Neandertal HLA-A and HLA-C for their
current distribution worldwide. All four combi-
nations have highest frequencies in Eurasia and
are absent in Africa (Fig. 3, B and C, and fig. S19).
Such conservation and distribution strongly sup-
port introgression of these haplotypes into mod-
ern humans by admixture with Neandertals in
Eurasia. The Neandertal HLA-B and -C alleles were
sufficiently resolved for us to study their distribution
in modern human populations (fig. S20); their
frequencies are high in Eurasia and low in Africa
(Fig. 3, D to G, and fig. S21). Our simulations of
frequency and haplotype diversity in Eurasia that
we observed (Fig. 1 and fig. S11) and was particu-
larly strong for B*51 and C*07:02 (fig. S22), and
the presence of such alleles in Africa was due to
back-migrations. Thus, Neandertal admixture con-
tributed B*07-, B*51-, C*07:02-, and C*16:02-
bearing haplotypes to modern humans and was
likely the sole source of these allele groups. Un-
like the distributions of Denisovan alleles, which
center in Asia (Fig. 2, E to G), Neandertal alleles
display broader distributions peaking in different
regions of Eurasia (Fig. 3, D to G).
Modern populations with substantial levels of
archaic ancestry are predicted to have decreased
LD (23). From analysis of HapMap populations
(20), we find that HLA class I recombination rates
are greater in Europeans (1.7- to 2.5-fold) and
Asians (2.9- to 7.7-fold) than in Africans, con-
REPORTS
HLA class I alleles (Fig. 4A). Enhanced LD de-
cay correlates with the presence of archaic alleles
(Fig. 4B and fig. S23), and the strongest correla-
tion was with HLA-A, for which the six haplo-
types exhibiting enhanced LD decay are restricted
to non-Africans. These haplotypes include A*24:02
and A*31:01, along with the four archaic allele
groups we characterized (A*11, A*26, and two
A*02 groups). A*24:02 and A*31:01 are common
in non-Africans and thus likely also introgressed
from archaic to modern humans. From the com-
bined frequencies of these six alleles, we estimate
the putative archaic HLA-A ancestry to be >50%
in Europe, >70% in Asia, and >95% in parts of
PNG (Fig. 4, C and D). These estimates for HLA-A
are much higher than the genome-wide estimates
of introgression (1 to 6%), which shows how
limited interbreeding with archaic humans has,

Downloaded from www.sciencemag.org on February 28, 2013

HLA allele introgression predicted the increased sistent with their higher frequencies of archaic in combination with natural selection, substan-
tially shaped the HLA system in modern human
populations outside of Africa. Our results demon-
A B strate how highly polymorphic HLA genes can
Recombination Population be sensitive probes of introgression, and we pre-
HLA-A
Genomic African European Chinese Japanese
Population rate (cM/Mb) dict that the same will apply to other polymorphic
region Distinct alleles 22 17 14 11
Mean SD Number 0 5 3 5
immune-system genes, for example, those encod-
Allele-specific ing the killer-cell immunoglobulin-like receptors
African 0.33 0.02 02:01 02:01 02:01
haplotypes Defining
HLA-A-B European 0.57 0.05
02:06 02:06 (KIR) of NK cells. Present in the Denisovan ge-
allele ---
(~1.4Mb) Japanese 0.98 0.11
with
---
11:01 11:01 nome (11), a candidate KIR for introgression is
name 24:02 24:02 24:02
Chinese 1.07 0.11
enhanced
26:01 26:01 KIR3DS1*013 (Fig. 4E), rare in sub-Saharan Af-
LD decay ( A* )
African 0.32 0.03
--- 31:01 ricans, but the most common KIR3DL1/S1 allele
HLA-A-C European 0.57 0.06 C outside Africa (24).
(~1.3Mb) Japanese 0.92 0.09 On migrating out of Africa, modern humans
Population Putative archaic ancestry at HLA-A
Chinese 0.95 0.10 encountered archaic humans, residents of Eurasia
African 6.7%
African 0.51 0.10 for more than 200,000 years, who had immune
European 51.7%
HLA-C-B European 1.72 0.39 systems better adapted to local pathogens (25).
Chinese 72.2%
(~85kb) Japanese 2.52 0.52 Japanese 80.7% Such adaptations almost certainly involved changes
Chinese 3.94 0.81 Papua New Guinea 82.3% [65.9% - 95.3%] in HLA class I, as exemplified by the modern hu-
man populations who first colonized the Americas
D E
(17, 18). For small migrating populations, admix-
ture with archaic humans could restore HLA di-
versity after population bottleneck and also provide
a rapid way to acquire new, advantageous HLA
variants already adapted to local pathogens. For
example, HLA-A*11, an abundant archaic allo-
type in modern Asian populations, provides T
cellmediated protection against some strains of
Epstein-Barr virus (EBV) (26), and in combina-
98.2
90.0 52.3 tion with a peptide derived from EBV, is one of
45.0
60.0 30.0 only two HLA ligands for the KIR3DL2 NK cell
30.0 af af
15.0 (%) receptor (27). HLA*A11 is also the strongest lig-
(%)
0.0 0.0
and for KIR2DS4 (28). Other prominent intro-
Putative archaic ancestry at HLA-A KIR3DS1*013 gressed HLA class I proteins are good KIR ligands.
HLA-B*73 is one of only two HLA-B allotypes
Fig. 4. LD decay patterns of modern HLA haplotypes identify putative archaic HLA alleles. (A) HLA class I
carrying the C1 epitope, the ligand for KIR2DL3
recombination rates in Eurasia exceed those observed in Africa. We focused on the three intergenic
regions between HLA-A, -B, and -C (left-most column) in the four HapMap populations (center column) (29). Prominent in Amerindians, C*07:02 is a
(20). Recombination rates were corrected for effective population size (11). (B) Enhanced HLA class I LD strong C1 ligand for KIR2DL2/3 and both B*51
decay significantly correlates with archaic ancestry (a = 0.0042) (11). Shown for each HapMap and A*24 are strong Bw4 ligands for KIR3DL1
population are (top row) the number of distinct HLA-A alleles present and (second row) the number (30). Such properties suggest that adaptive intro-
exhibiting enhanced LD decay [all allele-defining SNPs (correlation coefficient r2 > 0.2) are within 500 gression of these HLA alleles was driven by their
kb of HLA-A (31)]. The allele names are listed (rows 3 to 8) and colored green when observed in archaic role in controlling NK cells, lymphocytes essential
humans (Figs. 2 and 3) or associated with archaic-origin haplotypes (fig. S25). HLA-B and -C are shown in for immune defense and reproduction (6). Con-
fig. S23. Dashed line indicates absent in the population. (C) Predicted archaic ancestry at HLA-A [on the versely, adaptive introgression of HLA-A*26,
basis of the six alleles of panel (B)] for the four HapMap populations and six populations from PNG; for the -A*31, and -B*07, which are not KIR ligands,
latter, mean and extreme values are given. (D and E) Worldwide distribution in modern human populations was likely driven by their role in T cell immunity.
of putative archaic HLA-A alleles (D) and KIR3DS1*013, a putative archaic NK cell receptor (E). Adaptive introgression provides a mechanism for

www.sciencemag.org SCIENCE VOL 334 7 OCTOBER 2011 93

REPORTS
rapid evolution, a signature property of the extra- 17. M. P. Belich et al., Nature 357, 326 (1992). Institute as part of the gorilla reference genome
ordinarily plastic interactions between MHC class 18. D. I. Watkins et al., Nature 357, 329 (1992). sequencing project. These data can be obtained from
19. M. L. Petzl-Erler, R. Luz, V. S. Sotomaior, Tissue Antigens the National Center for Biotechnology Information (NCBI)
I ligands and lymphocyte receptors (6). 41, 227 (1993). Trace Archive (www.ncbi.nlm.nih.gov/Traces). We also
20. The International HapMap Consortium, Nature 437, used reads generated by Washington University School of
References and Notes 1299 (2005). Medicine; these data were produced by the Genome
1. A. Gibbons, Science 331, 392 (2011). Institute at Washington University School of Medicine in
21. F. Cruciani et al., Am. J. Hum. Genet. 70, 1197 (2002).
2. V. Yotova et al., Mol. Biol. Evol. 28, 1957 (2011). St. Louis and can be obtained from the NCBI Trace
22. Y. Moodley et al., Science 323, 527 (2009).
3. R. E. Green et al., Science 328, 710 (2010). Archive (www.ncbi.nlm.nih.gov/Traces/). Funded by NIH
23. M. DeGiorgio, M. Jakobsson, N. A. Rosenberg, Proc. Natl.
4. D. Reich et al., Nature 468, 1053 (2010). grant AI031168, Yerkes Center base grant RR000165,
Acad. Sci. U.S.A. 106, 16057 (2009).
5. V. Castric, J. Bechsgaard, M. H. Schierup, X. Vekemans, 24. P. J. Norman et al., Nat. Genet. 39, 1092 (2007). NSF awards (CNS-0619926, TG-DBS100006), by federal
PLoS Genet. 4, e1000168 (2008). 25. A. Ferrer-Admetlla et al., J. Immunol. 181, 1315 funds from the National Cancer Institute (NCI), NIH
6. P. Parham, Nat. Rev. Immunol. 5, 201 (2005). (2008). (contract HHSN261200800001E), and by the Intramural
7. K. Cao et al., Hum. Immunol. 62, 1009 (2001). 26. P. O. de Campos-Lima et al., Science 260, 98 (1993). Research Program of the NCI, NIH, Center for Cancer
8. P. Parham et al., Tissue Antigens 43, 302 (1994). 27. P. Hansasuta et al., Eur. J. Immunol. 34, 1673 (2004). Research. The content of this publication does not
9. C. Vilches, R. Pablo, M. J. Herrero, M. E. Moreno, 28. T. Graef et al., J. Exp. Med. 206, 2557 (2009). necessarily reflect the views or policies of the Department
M. Kreisler, Immunogenetics 40, 166 (1994). 29. A. K. Moesta et al., J. Immunol. 180, 3969 (2008). of Health and Human Services, nor does mention of
10. J. Robinson et al., Nucleic Acids Res. 39 (Database issue), 30. M. Yawata et al., Blood 112, 2369 (2008). trade names, commercial products, or organizations
D1171 (2011). 31. P. I. de Bakker et al., Nat. Genet. 38, 1166 (2006). imply endorsement by the U.S. government. Sequence
11. Materials and methods are available as supporting Acknowledgments: We thank individual investigators and the data have been deposited in GenBank under accession
material on Science Online. Bone Marrow Donors Worldwide (BMDW) organization nos. JF974053 to 70.
12. F. F. Gonzalez-Galarza, S. Christmas, D. Middleton, for kindly providing HLA class I typing data, as well as

Downloaded from www.sciencemag.org on February 28, 2013

A. R. Jones, Nucleic Acids Res. 39 (Database issue), D913
(2011).
13. C. Vilches, R. Pablo, M. J. Herrero, M. E. Moreno,
M. Kreisler, Immunogenetics 40, 313 (1994).
14. D. M. Behar et al., Am. J. Hum. Genet. 82, 1130 (2008).
15. O. Semino, A. S. Santachiara-Benerecetti, F. Falaschi,
L. L. Cavalli-Sforza, P. A. Underhill, Am. J. Hum. Genet.
70, 265 (2002).
16. M. J. Jobin, J. L. Mountain, Bioinformatics 24, 2936 (2008).
bone marrow registries from Australia, Austria, Belgium, Supporting Online Material
Canada, Cyprus, Czech Republic, France, Ireland, Israel, www.sciencemag.org/cgi/content/full/science.1209202/DC1
Italy, Lithuania, Norway, Poland, Portugal, Singapore, Materials and Methods
Spain, Sweden, Switzerland, Turkey, United Kingdom, Figs. S1 to S26
and United States for contributing typing data through References (3287)
BMDW. We thank E. Watkin for technical support. We are
indebted to the large genome sequencing centers for 1 June 2011; accepted 5 August 2011
early access to the gorilla genome data. We used Published online 25 August 2011;
sequence reads generated at the Wellcome Trust Sanger 10.1126/science.1209202

Aboriginal Australians are predicted to have di-

An Aboriginal Australian Genome versified from within the Asian cluster [for defi-
nitions of human populations and groups, see (4)]
Reveals Separate Human (Fig. 1A, top). Recent whole-genome studies re-
veal a split between Europeans and Asians dat-
Dispersals into Asia ing to 17,000 to 43,000 years before the present
(B.P.) (5, 6). Because greater Australia (Australia
and Melanesia, including New Guinea) has some
Morten Rasmussen,1,2* Xiaosen Guo,2,3* Yong Wang,4* Kirk E. Lohmueller,4* Simon Rasmussen,5 of the earliest archaeological evidence of ana-
Anders Albrechtsen,6 Line Skotte,6 Stinus Lindgreen,1,6 Mait Metspalu,7 Thibaut Jombart,8 tomically modern humans outside Africa, dating
Toomas Kivisild,9 Weiwei Zhai,10 Anders Eriksson,11 Andrea Manica,11 Ludovic Orlando,1 back to ~50,000 years B.P. (7, 8), a divergence of
Francisco M. De La Vega,12 Silvana Tridico,13 Ene Metspalu,7 Kasper Nielsen,5 Mara C. vila-Arcos,1 aboriginal Australasians from within the Asian
J. Vctor Moreno-Mayar,1,14 Craig Muller,15 Joe Dortch,16 M. Thomas P. Gilbert,1,2 Ole Lund,5 cluster is not compatible with population conti-
Agata Wesolowska,5 Monika Karmin,7 Lucy A. Weinert,8 Bo Wang,3 Jun Li,3 Shuaishuai Tai,3 nuity in Australia. Alternatively, on the basis of
Fei Xiao,3 Tsunehiko Hanihara,17 George van Driem,18 Aashish R. Jha,19 Franois-Xavier Ricaut,20 archaeological and fossil evidence, it has been
Peter de Knijff,21 Andrea B Migliano,9,22 Irene Gallego Romero,19 Karsten Kristiansen,2,3,6 proposed that greater Australia was occupied by
David M. Lambert,23 Sren Brunak,5,24 Peter Forster,25,26 Bernd Brinkmann,26 Olaf Nehlich,27 an early, possibly independent out-of-Africa dis-
Michael Bunce,13 Michael Richards,27,28 Ramneek Gupta,5 Carlos D. Bustamante,12 persal, before the population expansion giving
Anders Krogh,1,6 Robert A. Foley,9 Marta M. Lahr,9 Francois Balloux,8 Thomas Sicheritz-Pontn,5,29 rise to the majority of present-day Eurasians (9, 10).
Richard Villems,7,30 Rasmus Nielsen,4,6 Jun Wang,2,3,6,31 Eske Willerslev1,2 According to this multiple-dispersal model,
the descendants of the earlier migration became
We present an Aboriginal Australian genomic sequence obtained from a 100-year-old lock of assimilated or replaced by the later-dispersing
hair donated by an Aboriginal man from southern Western Australia in the early 20th century. populations, with a few exceptions that include
We detect no evidence of European admixture and estimate contamination levels to be below Aboriginal Australians (10, 11) (Fig. 1A, bottom).
0.5%. We show that Aboriginal Australians are descendants of an early human dispersal into We sequenced the genome of an Aboriginal
eastern Asia, possibly 62,000 to 75,000 years ago. This dispersal is separate from the one that Australian male from the early 20th century to
gave rise to modern Asians 25,000 to 38,000 years ago. We also find evidence of gene flow overcome problems of recent European admix-
between populations of the two dispersal waves prior to the divergence of Native Americans ture and contamination (4). We used 0.6 g of hair
from modern Asian ancestors. Our findings support the hypothesis that present-day Aboriginal for DNA extraction (4, 12). Despite its relatively
Australians descend from the earliest humans to occupy Australia, likely representing one of the young age, the genomic sequence showed a high
oldest continuous populations outside Africa. degree of fragmentation, with an average length
of 69 base pairs. The genome was sequenced to
he genetic history of Aboriginal Austra- mans out of Africa, followed by subsequent serial an overall depth of 6.4; the ~ 60% of the ge-

T lians is contentious but highly important

for understanding the evolution of modern
humans. All living non-African populations like-
ly derived from a single dispersal of modern hu-
founder effects (1, 2). Accordingly, eastern Asia
is hypothesized to have been populated by a
single early migration wave rather than multiple
nomic regions covered was sequenced to an av-
erage depth of 11 (4) [theoretical maximum is
~85% (12)]. Cytosine-to-thymine misincorpora-
dispersals (3). In this single-dispersal model, tion levels typical of ancient DNA (13) were low

94 7 OCTOBER 2011 VOL 334 SCIENCE www.sciencemag.org